电力信息与通信技术2025，Vol.23Issue(5)：77-83,7.DOI:10.16543/j.2095-641x.electric.power.ict.2025.05.10
基于掩码自编码器的智能电网虚假数据注入攻击检测方法
A Method of Detecting False Data Injection Attacks in Smart Grid Based on Masked Autoencoder
摘要
Abstract
False data injection attacks typically consist of two steps:system intrusion and data manipulation.Existing research predominantly focuses on detecting data manipulation,whereas studies on detecting false data injection attacks through the identification of system intrusions are relatively scarce.To address this gap,this paper proposes a method based on a masked autoencoder.Initially,the method utilizes statistical features and raw byte features of communication flows in smart grids,combining a feature fusion mechanism to transform communication flows into feature grayscale images.Subsequently,leveraging an enhanced MAE,the approach detects the attack traffic of false data injection attacks intruding into the system from the perspective of computer vision.Experimental results demonstrate that the proposed method achieves efficient detection of false data injection attacks with up to 99%accuracy across multiple datasets.关键词
虚假数据注入攻击/流量分类/掩码图像预测Key words
false data injection attack/traffic classification/masked image prediction分类
动力与电气工程引用本文复制引用
刘佳羽,尚涛,姜亚彤,熊科宇..基于掩码自编码器的智能电网虚假数据注入攻击检测方法[J].电力信息与通信技术,2025,23(5):77-83,7.基金项目
国家电网有限公司总部科技项目资助"电力监控系统网络安全威胁综合管控及过程推演关键技术研究"(5108-202040036A-0-0-00). （5108-202040036A-0-0-00）
